You might want to grow multiple seeds from the hybrid. I would expect if you grow enough of them, at least one of them should result in elimination of the unsavory poncirus taste. That is assuming that the genes behave as normal dominant/recessive pairs. I believe how gene expression works in citrus is a little more complicated.
Let B represent bad taste, and b represent lack of bad taste.
Also let C represent cold hardiness and c represent lack of cold hardiness. B is dominant over b which is recessive.
the first cross was the poncirus with the pure pomelo
BB CC x bb cc = Bb Cc
next you obtained a zygotic seed from the resulting previous hybrid
Bb Cc x Bb Cc =
these are the statistical probabilities:
BB CC
BB Cc
BB cC
BB cc
Bb CC
Bb Cc
Bb cC
Bb cc
bB CC
bB Cc
bB cC
bB cc
bb CC
bb Cc
bb cC
bb cc
only the last 4 possibilities will not have the bad poncirus taste. 4 out of 12 is a 1 in 4 probability. Out of that 1 in 4, 1 in 4 will not really have any cold hardiness. Half will have moderate cold hardiness, like the hybrid parent, and the remaining 1 out of 4 will have superior cold hardiness, better than the parent. That means there will be a 1 out of 16 chance the offspring (assuming they are all zygotic) will both lack a bad taste and also be more cold hardy than the first generation hybrid.
This is probably oversimplified, of course.
what you have now would probably be more like a Bb CC type.
I apologize for hijacking your thread if theory is not what you wanted the discussion to focus on.